IAC Summary on New Drugs and New Combinations – How will They Change What We Do?
Vienna July 18-23 2010
Joe Eron, Univ of North Carolina
This year’s IAC was one of the most exciting and data-filled International AIDS meetings since Vancouver in 1996. The defining presentation of the meeting was the CAPRISA 004 study that demonstrated that a topical microbicide (tenofovir gel) prevented HIV infection in at-risk women in South Africa. The excitement in the room during the series of presentations was palpable and the presentations were interrupted several times by applause. Much as Vancouver signaled the beginning of the HAART era, perhaps CAPRISA 004 will signal the practical beginning of the Antiretroviral Prevention era. Certainly a prevention intervention that can be controlled by at-risk women is a true breakthrough. The hallway discussions were about the magnitude of the effect (Did we expect more? Did we expect less?), the obvious importance of adherence and perhaps the challenging questions about whether a second confirmatory study was needed before this intervention could be made available to women and whether placebo arms should still be incorporated in further studies of microbicides and other pre-exposure prophylaxis (PrEP) studies in women.
While CAPRISA 004 was a clear centerpiece of the meeting there were also several presentations of antiretroviral therapies (used as treatment in the old fashioned way) that were quite important and will certainly impact how we treat our patients over the next 2-3 years.
ECHO and THRIVE
The most important of these presentations was made by Cal Cohen who presented the Phase III data comparing rilpivirine head-to-head with efavirenz when both were combined with two NRTI (predominantly tenofovir DF/FTC fixed dose combination) in treatment naïve patients. These data are the most clinically relevant because this was one of only two presentations of phase III data in treatment naive patients, because it looks like rilpivirine is ready to be combined with TDF/FTC in a fixed dose combination and most importantly, because Tibotec has submitted the new drug application (NDA) for this drug on July 28th 2010. The full details of Dr. Cohen presentation can be found in the detailed NATAP summary but there are several interesting nuances to the results that we will probably be discussing and debating for some time. First and foremost in the primary analyses of both studies, which were independent well powered studies (EHCO used FDC TDF/FTC as the NRTI backbone in all patients, whereas THRIVE allowed the use of TDF/FTC, ZDV/3TC or ABC/3TC as the backbone though the majority of the participants received TDF/FTC), rilpivirine with 2 NRTI was non-inferior to EFV and 2 NRTI in intention-to-treat analyzes (VL <50 c/mL, ITT-TLOVR at Week 48). Both arms in each study performed very well in a challenging population (median baseline viral load was 100,000 c/mL). In the analysis that combined the results of the 2 studies 84.3% of participants on RPV-based therapy had HIV RNA < 50 copies/mL at 48 weeks compared to 82.3% of participants on the EFV-based therapy. The “non-inferiority” of RPV was well established with the lower bound of the confidence interval at -2.2%, way below the non-inferiority bound of -12%. Though I am sure this is not FDA-kosher you could almost call the two therapies (when viewed in the light of the combined results) “equivalent”. However ….. RPV-based and EFV-based therapy produced such similar overall results for different reasons.
Bioequivalence of the Co-Formulation of Emtricitabine/Rilpivirine/Tenofovir DF – (08/04/10)
RPV-based therapy was clearly better tolerated than EFV-based therapy. In the results from the 2 studies combined, Grade 2-4 AE that were at least possibly related to study medication were twice as likely on EFV-based therapy (31% vs. 16%, p < 0.0001) and significantly fewer participants receiving RPV-based therapy discontinued study medication compared to those on EFV-based therapy (3% vs. 8% p = 0.0005). These discontinuations of course are counted as failures in the TLOVR-ITT analysis. As anticipated neurologic and psychiatric adverse events were significantly and substantially less common with RPV and unanticipated (at least by me) was that rash was also significantly less common (3% vs. 14% p < 0.0001). Grade 3-4 laboratory abnormalities were also significantly less common in RPV-treated participants. While no single grade 3-4 lab abnormality was very common in either arm, grade 3-4 increases in ALT, LDL cholesterol, total cholesterol and triglycerides were all less common with RPV. Mean change in all individual lipid parameters (including HDL) were greater with EFV with very little change in any of the parameters in those participants on RPV-based therapy. Because HDL also rises on EFV-based therapy the total cholesterol to HDL ratios at week 48 were not significantly different between the two treatment groups in the combined analysis.
The flip side to the superior tolerability data for RPV-based therapy was that virologic failures (as defined in the TLOVR analysis) were approximately twice as common in participants on RPV (9.0% vs. 4.8% in the combined analysis, p value not reported but likely very significant). So in effect, the discontinuations due to AE (approximately a 5% difference) are balanced by VF (approximately a 4% difference). There were some relatively small but curious differences in VF between the two studies with a greater difference in VF in the ECHO study (6.6%), where all participants received TDF/FTC, compared to THRIVE (1.8%), where there was a choice of NRTI. We were told that response rates were not different in THRIVE by NRTI choice so we don’t really have an explanation for this difference in virologic failures. Perhaps the participants in THRIVE were more likely to take their medications more consistently (the difference in discontinuations due to AE between the two arms was also somewhat less in THRIVE).
The resistance consequences were also different between the two treatment groups as one might expect. Sixty-two participants out of 686 on RPV had virologic failure with resistance data available compared to 28 out of 682 participants on EFV-based therapy. In these participants with VF and resistance data, the likelihood of acquiring at least one mutation was greater in RPV recipients (71% vs. 57%) and while the proportion with NNRTI resistance was similar (63% vs. 54%), NRTI mutations were more common in RPV failures (68% vs. 32%). The most frequent NNRTI RAM in VF with resistance assays on the EFV arms was the well known K103N mutation. However in participants with VF on the RPV arms the most common NNRTI resistance mutation was E138K. Preliminary phenotypic information demonstrated that virus that developed resistance to RPV were likely to be cross resistant to etravirine.
These data suggest that there may be a subtle difference in potency between efavirenz and RPV given at the dose chosen for the Phase III studies. Recall that higher doses of RPV (while well tolerated) were associated with QT prolongation on ECG. Therefore doses of 75 mg and 150 mg daily (as opposed to the dose of 25 mg daily that was chosen) were not acceptable for the Phase III studies. If we examine the results in the sub-groups with viral load < 100,000 and > 100,000 we find support for the potency argument. In the participants with lower viral loads (VL) RPV-based therapy performed very well as the tolerability advantage held sway. In fact, in the analysis of the combined study RPV-based therapy was actually superior to EFV-based therapy as the potential subtle difference in antiviral activity was not a disadvantage in the lower viral load group. In higher viral load group the overall result was that RPV was still non-inferior though the proportion responding was numerically less in the RPV-treated participants. The hypothesis is that the tolerability advantage is somewhat neutralized by more virologic failures. The differences between the two trials also show up here. In THRIVE there is little difference in overall response in the high viral load groups between RPV and EFV (79% vs 80% < 50 c/mL at 48 weeks by ITT TLOVR), whereas in ECHO the results were 76% and 82% respectively. There was no obvious explanation for this difference between studies but one could imagine if there were modest difference in drug exposure (perhaps due to adherence) in the two studies that might explain the results. If RPV is more on the cusp of its concentration to activity ratio then small differences in exposure or very high viral loads might result in a greater risk of virologic failure. In contrast the response rates on efavirenz were very consistent across the two studies and in the lower and higher viral load groups. This result is very consistent with our experience with EFV-based therapy in almost all other studies. For now though my speculation is only one possible hypothesis, which perhaps will be confirmed with more detailed analyses that include drug concentration measurements and further sub-analyses of the two studies.
The results of ECHO and THRIVE and the combined analyses set up an interesting discussion about how we will use EFV/TDF/FTC and RPV/TDF/FTC if RPV (and perhaps the fixed dose combination) is approved in early 2011, which seems very likely. At some point next year we will likely have two “one pill once a day” therapies. There is a clear tolerability advantage to RPV, which is a substantial plus as we move to treating asymptomatic patients with higher and higher CD4 cell counts. On the other hand there will be concerns about the increased risk of virologic failure that may be more likely in patients who have higher baseline viral loads. Again the sequencing argument will be raised. If you have virologic failure on RPV-based therapy it appears likely that the patient’s virus will also be etravirine resistant, whereas if you have virologic failure on EFV based therapy and the only mutation is K103N then etravirine will still have full activity. We hate losing a drug, even if we weren’t planning to use it any time soon. Alternatively the loss of etravirine in a small fraction of patients who start RPV-based treatment will have very little practical consequence as there are multiple antiretroviral combinations available for first line failures. There will also be discussions of the better lipid effects of RPV, however we don’t have the very long term safety data that we have with EFV. RPV has NOT shown any teratogenicity in lab based and preclinical studies and will likely be a category C medication. Many care providers and HIV-infected women of child bearing potential may then prefer RPV based therapy. Another potential option will be to switch to RPV-based therapy (especially when the combination pill is approved) once patients are suppressed. Obviously this strategy has not yet been studied but I suspect it will be, once rilpivirine is available. There also will be a lot more data forth coming from the ECHO and THRIVE studies including new sub-analyses and eventually the 96 week data. While I don’t think ECHO and THRIVE can be considered a “home run” I think we all look forward to having rilpivirine and another one pill once a day therapy available to us soon.
THE “ING” STUDIES
We learned a lot more about another promising drug, the Shionogi GSK integrase inhibitor S/GSK 1349572. This drug, “572”, was introduced last summer at the IAS meeting in Cape Town and was shown to be very potent in a short term monotherapy study in integrase inhibitor naïve subjects reducing plasma HIV RNA by over 2 log10 over 10 days. In Vienna we saw longer term phase IIb data in treatment naïve individuals and we saw the initial activity of this drug in raltegravir failures who had documented integrase inhibitor resistance.
The treatment naïve study, called SPRING, was a partially blinded phase IIb 48 week study examining 3 different doses of 572 (10, 25 and 50 mg) with a comparator arm of efavirenz each combined with 2 NRTI (approx 2/3 TDF/FTC and 1/3 ABC/3TC). Sixteen week data were shown and demonstrated very rapid suppression (consistent with previous raltegravir and elvitegravir data) in each of the three 572 arms with over 80% of patients suppressed to < 50 c/mL by week 8 and 90% or more < 50 c/mL at week 16 in each of the 3 arms. Again consistent with raltegravir data the speed of suppression was significantly faster with 572 compared to efavirenz-based though it is likely that over 48 weeks the treatment arms will look similar in the relatively small study of 50 participants per arm. 572 was very well tolerated and there were no serious adverse events related to 572 and very few discontinuation in any of the arms. The highest dose (50 mg once daily) was chosen for further development though the 3 572 dose arms in this study will be continued through 48 weeks as there are little long term data with this agent. The potential advantages of 572 in naïve patients over other integrase inhibitors include; once daily dosing that does not require a boosting agent, a low mg dose that will allow co-formulation and probable single pill therapy and a predictable concentration activity relationship (pharmaco-dynamic profile). The drug seems very well tolerated over the short term, like other integrase inhibitors, which is likely to be an advantage even when comparing across class, though clearly longer term tolerability and toxicity data are needed as well as data on lipid changes and other laboratory markers. Based on the SPRING results the naïve treatment market is likely to get more crowded with multiple good choices over the next several years.
I had the good fortune to present the VIKING study, which examined the short term activity of 572 in patients with documented RAL resistance. The results of this study were interesting and more complex than the very straightforward results of the SPRING study. In this study participants had to have documented RAL resistance at screening and could either be on RAL-based combination therapy or combination therapy without RAL, having failed RAL in the past with persistent resistance. Based on in vitro susceptibility data there were 2 groups of subjects. One group had integrase resistance mutations at screening at codon 155 with or without others, 143 again with or without other mutations or a single mutation at the 148 codon. The second group had to have a mutation at the 148 codon but then also had one or more secondary mutations. In vitro most viruses with the 155 pattern or the 143 pattern remain fully or nearly fully susceptible to 572. In contrast viruses that have the 148 mutation plus additional mutations (usually 140S with or without others) have decreased susceptibility to 572, though the fold change of some of these viruses are in a range where 572 might still have activity. All participants received 572 functional monotherapy for 10 days at 50 mg daily. Those who were on RAL (the majority) had RAL replaced by 572 with continued background. Those who had already stopped RAL and were on a stable combination regimen had 572 added for the 10 day period. On day 11 participants were allowed to optimize their background therapy and continue the 572. Only the functional monotherapy data were presented.
Twenty-seven subjects were enrolled, 18 in group 1 and 9 in group 2. At baseline the median CD4 cell count was 110 and median viral load was approximately 30,000 c/mL. These were highly treatment experienced patients. They had received a median of 17 antiretroviral agents in the past and the majority had been exposed to darunavir, enfuvirtide and etravirine. The median baseline fold change to RAL was 161 while the median baseline fold change to 572 was 1.5. There was a clear relationship between the pattern of RAL resistance mutations and the baseline fold change with the highest fold changes seen when the codon 148 mutation was present with 2 or more secondary mutations. Viruses with the 143 mutation (with or without other mutations) had minimal if any shift in susceptibility to 572. The primary endpoint of the study was at least a 0.7 log10decline in HIV RNA from baseline or a viral load < 400 c/mL at day 11. Response rates were consistent with the in vitro data and the baseline genotype and phenotype data. All 18 participants in group 1 (virus with 155 or 143 complex mutations) met the primary endpoint with a mean change in HIV RNA of 1.82 log10. Eight out of 18 had HIV RNA < 400 c/mL at day 11 and one was < 50 c/mL. In contrast only 3 of the 9 subjects in group 2 with 148 plus at least one additional mutation met the primary endpoint, though all 3 did fall to < 400 c/mL. The mean change in this group was 0.72 log10. Overall there was a very strong inverse correlation between baseline fold change and the change in HIV RNA at day 11. Importantly no participant had emergence of a new RAL resistance mutation over the 11 day period. In 17/18 subjects in whom baseline and day 11 resistance testing could be accomplished the change in susceptibility was < 2 fold. A detailed resistance poster was presented by Bonaventura Clotet. He showed that the one subject who had a an approximate 6 fold decrease in susceptibility had a mixture of viruses at baseline and over the 11 day period the virus with the more resistance genotype (148 plus an additional mutation) became the predominant variant and likely accounts for the decrease in susceptibility that occurred over the 11 days of functional monotherapy.
Following the presentation there was a lot of discussion about the results and whether 572 fits the bill as a second generation integrase inhibitor. Clearly some viruses with substantial resistance to RAL respond very well to 572. The viruses with the 155 and 143 mutation patterns were not just early failures of RAL. Though the duration of previous RAL therapy for these subjects is known (though not reported in the presentation by Eron et al) the duration of RAL failure for each patient is not known. However, as was shown in the poster by Dr. Clotet, many of the viruses with the 155 and the 143 pathways had multiple mutations and had substantial resistance to RAL. Participants with these viruses have very good responses over the short term and if adequate partner therapy can be given the response is likely to be durable. The flip side is that the 148 plus 140 pathway is probably the most common resistance pattern with RAL failures especially if virologic failure has been ongoing for some time in the presence of RAL. The 155H mutation may show up early but in some patients the virus undergoes a transition to the 148/140 path. These viruses did not respond as well to 572. There was a decrease in HIV RNA in most of the patients that had 148 plus one or more mutations but the effect was modest (mean of 0.72 and an approximate median of about 0.5 log10). 572 at the 50 mg once a day dose is not likely to have a major impact in individuals with these viruses unless there were several other active agents that could be combined in an optimized regimen. The logical question (which was asked by Bernard Hirschel) was what about a higher dose. 572 has been well tolerated and the dose response curves that have been generated would suggest that a higher dose should have a greater activity. The VIKING study is now enrolling a cohort 2 – which will test 50 mg twice daily. Hopefully this second cohort will show us whether more robust responses can be seen again viruses that have moved down the 148 pathway. An additional clinical point that could be taken from the VIKING results are that individuals who are on raltegravir and who are failing should probably be moved to an alternative regimen or have their raltegravir stopped if at all possible. Prolonged failure in the face of raltegravir selective pressure may push the virus to greater cross-resistance to 572.
The last topic I want to touch on just briefly is the topic on NRTI-sparing regimens, in particular for treatment naïve patients. It had never really occurred to me but if you look at the DHHS guidelines there is not one single NRTI-sparing first-line regimen listed in any category (Preferred, Alternative, Acceptable, possibly Acceptable). One could argue that perhaps we just don’t need NRTI-sparing regimens in first line therapy, but as our patient’s age and the very long term toxicity of nucleosides becomes apparent there is great pressure to keep looking.
IAC produced several pilot studies of NRTI-sparing regimens. The largest, called PROGRESS, compared LPV/r plus RAL both BID with LPV/r plus TDF/FTC over 48 weeks in over 200 subjects. The bottom line is that the 2 treatment arms looked very similar virologically and there were very few virologic failures with resistance in either arm. A strong word of caution though is that the baseline viral load was very low, less than 20,000 c/mL (remember ECHO and THRIVE had a median baseline VL of 100,000). LPV/r plus RAL may be a very good NRTI sparing regimen but it needs to be challenged in patients with higher viral load. In addition, the lipid levels were higher on the NRTI-sparing arm probably reflecting the lipid lowering effect of TDF. In this modest sized short term study there were not major differences in toxicity.
The small pilot SPARTAN study which combined RAL and atazanavir (ATV) unboosted both BID provides a contrast to PROGRESS. While the overall results were similar in this arm to ATV/r plus TDF/FTC at 24 weeks (2:1 randomization with 63 vs. 30 subjects in each arm), there were more subjects on RAL plus ATV who had virologic failure with resistance testing results. There were 11 VL failures (VL > 50 at week 24) on the RAL arm, 8 had baseline VL greater than 250,000 c/mL, 6 had HIV RNA > 400 copies and had resistance testing and 4 had RAL resistance mutations. In contrast, while there were 8 “virologic failures” on the ATV/r TDF/FTC arm only 1 had a viral load greater than 400 and that patient had no resistance mutations. Bilirubin levels were also higher on the RAL/ATV arm and there was no suggestion that ATV concentrations on that arm were sub-optimal. I think we need more data on the RAL plus PI (preferably boosted) in high viral load patients before we venture to far down this path in the clinic, despite the PROGRESS results.
One other NRTI-sparing regimen that was tested was maraviroc (150 mg once daily) plus ATV/r compared to a reference arm of ATV/r plus TDF/FTC. Again this was a small study with 60 patients per arm. At 24 weeks a greater proportion of patients on the NRTI arm were less than 50 c/ml (89%) though this is a very high number for ATV/r TDF/FTC – much higher than was seen in the CASTLE study for example. MVC plus ATV/r had 80% < 50 c/mL at 24 weeks so perhaps this may be an NRTI sparing regimen to test in a larger number of patients. The number of subjects with baseline VL > 100,000 in this study was small but MVC plus ATV/r seems to do similarly in low and high VL groups.
I don’t think NRTI-sparing is ready for clinic prime time on a large scale but perhaps there are situations where it is appropriate and again switching once suppressed may be a safer approach but to date we only have pilot type data in this setting too. A large trial of DRV/r plus raltegravir compared to DRV/r plus TDF/FTC in treatment naïve patients is about to start in Europe and it should provide some answers – though we will have to wait for some time to see these results.